Opinion - AMD's actions in recent quarters reveal a trend we can only describe as: all in. It's a poker term meaning AMD bet it all on Barcelona. But is Barcelona enough? Can AMD make it with this product as their flagship?
This article is an opinion piece so I can admit that, personally, I'm a big fan of AMD's public vision. I have been for a long time as I believe in their goals and product ideas. They're forward thinking and fresh. In fact, I'd like to be where they want to go. I think AMD's stated vision is on track with what's needed and their product goals represent a good plan.
It's just that I see one overriding reality: A lack of cash.
AMD is looking to repay about $5.1 billion in debt in 2012. That's just 4.5 years away. They'll have to repay just under $1 billion per year if they average it out, or just under $250 million per quarter. Every quarter. Every year. That ongoing expenditure must be had to repay the $3.7 billion in senior convertible notes they sold earlier this year, plus the 5.75% annual interest. This debt will be in addition to AMD's planned expenditures between now and then, including ongoing expenses. In the alternative, AMD must repay all that debt in 2012 based on that year's annual revenue alone. That could account for half of their revenue that year.
AMD has plans for several new products and technologies in that time. They're expecting to see 45 nm production by the end of next year. 32 nm will come either late in 2009 or early in 2010. Neither of those moves come cheap. They're also developing two new lines of processors called Bobcat and Bulldozer. These will have new technologies and extensions which were described recently by AMD. They're still being developed and that's R&D expense. These new CPUs will also target new markets where AMD currently has no presence. That's new expansion and capital expenses. And AMD wants to convert Fab 30 into Fab 38, a 300 mm 65 nm SOI production facility like Fab 36.
In short, it all adds up to a lot of money over just a few years. So, just where is AMD going to get the cash to pay for all of it and their $250 million per quarter debt?
AMD's annual revenue for 2006 was $7.6 billion. They lost $47 million for the year. In 2012, AMD must pay back $5.1 billion in hard cash from an unknown annual growth rate beginning at this current base of $7.6 billion in 2006 and the significant losses we know of in 2007.
Even with significant growth, growth which will come with a heavy price in expansion, capacity and a cost of doing business, how will AMD will be able to do this? I ask AMD this question primarily because of what is being seen at Intel.
The current price war has hit AMD hard. AMD lost hundreds of millions in an attempt to keep market share high. They brought prices low and have paid the price. It's cost them over $500 million per quarter in the last two quarters. Today, their highest-end dual-socket Opteron (model 2350) sells for only $389 and their quad-socket 8350 is $1019.
AMD cannot compete in volume, which means they cannot compete in price (unless their is some magic in production efficiency I don't know about). They cannot compete in performance unless they solve their manufacturing issues. They've had six extra months during Barcelona's delay to do that, but with no success in bringing the clock speeds back up to their early expectations. We'll have to wait until after Intel's 45 nm launch for that as AMD has promised 2.5 GHz parts by the end of this year. Intel's 45nm launch is coming in two months (Nov 11), meaning AMD will not be able to compete in terms of performance or, very likely, performance-per-watt (due to Intel's 45nm advantage).
Barcelona does bring architectural advantages yielding better performance. But all of them are needed today just to compete against Intel's current products. Never mind the higher clock rates and lower power budgets we'll see very shortly from Intel at 45nm as Intel is not sitting idly by waiting for Core 2 to run its course. Intel has a 2-year strategy in effect and they have the cash to deliver on it. We've seen that in practice.
Intel will release over a dozen 45 nm products next year. They'll have three 45 nm full-time production fabs cranking out parts by the end of the year. Their 65 nm process technology is already very mature. In the benchmarks released today we saw Intel's current processors able to compete with Barcelona on nearly all fronts. And Barcelona's occasional win percentages were much smaller than Xeon's. All of this while Intel's 45nm processors, those which will compete with Barcelona, are on track for a Nov 11 launch.
Advanced process technology
Intel continues to develop its high-k/metal gate solution for production 45 nm parts. According to Intel's press releases, the high-k/metal gate technology has the potential of reducing gate leakage by 10x over previous implementations. That means lower power, less heat and greater clock speeds, all of which hit Barcelona right in its sweet spot.
AMD has no dedicated 45 nm fabrication facilities. All of their 45 nm test projects will happen at Fab36 in Dresden, Germany. It is also their only 300 mm fab and this will remain that way until Fab 30 is revamped into Fab38. And all of that revamping will come with significant expense, possibly a few billion.
Today, Intel's D1D research fab in Hillsboro, Oregon, serves as their first 300 mm 45nm production fab. Three more will go online by the end of 2008. One each in Arizona, Israel and New Mexico. Intel uses a "copy exactly" process for fabrication construction. This allows yields and equipment to be identical at each site. When they get a particular process technology working with yields and performance at rated goals, they exactly duplicate that fab someplace else. They literally duplicate everything exactly. Factors like humidity, temperature, wall paint, floor tiles, wiring, A/C ducts, electricity routing, wiring thickness, everything. It's all the same and it's truly a "copy exactly" process. This is what enables Intel to avoid surprises in yield.
Intel's manufacturing process can also mature so quickly and easily because any fabrication facility which finds a newer/better way to increase production or yields can immediately spread that knowledge to all other facilities. And because everything is the same everywhere else, the new process will implement and scale without unexpected events.
I see no way AMD can compete with these upcoming volumes. Even with supplemental manufacturing capacity by Chartered Semiconductor on their certified 65 nm SOI process. Intel is too large and has too much money.
AMD has mounting debt, consistent quarterly losses, unresolved manufacturing problems, a lack of significant cash on hand and they are capacity constrained. They're walking in uncharted waters with Fusion, Bulldozer, Bobcat and Barcelona is a less than stellar performer. And they're missing product deadlines while disclosing big future plans. All of this is happening at AMD while Intel attacks them from all sides with their tick tock strategy. Intel meets defined technology milestones and accomplishments, one right after the other. That's not a wish or desire, that's just reality.
AMD doesn't have much choice. They need something big to fall into their court. They need a favorable court ruling in their multi-jurisdiction anti-trust case against Intel. Or they need to be bought out by another company like IBM or Samsung. Without such a large windfall, I don't see any way AMD can compete with Intel's resources, capacity and technology for too much longer. The stakes are getting too high and the antes are too big.
AMD is all in. They've bet everything on Barcelona and unless something drastic happens I just don't see it living up.
What do I see happening in the future? I see the writing on the wall: I see Intel surpassing AMD in performance, manufacturing technology, capacity, execution, profitability and long-term product feasibility. I truly believe AMD has better goals for x86 than Intel does (at least those which are mapped out publicly) because AMD has listened to the market and brought us more desirable product visions. And for some of these, Intel even followed AMD's lead, like to x86-64. But at the end of the day it's about one thing: money. Intel has lots of it while AMD has lots of mounting debt.
In my opinion, Barcelona does not bring enough to the table to dig AMD out of their current hole. It will require an investment by Samsung or IBM to turn it around. Such a partnership or buyout could give AMD enough resources necessary for their products and visions to come true. Without it, I see Barcelona's launch as the beginning of the end for AMD.
These are my opinions and they come from much analysis and consideration. I welcome everyone to post their responses or opinions on this matter.